Towards Open Science: The Case for a Decentralized Autonomous Academic Endorsement System

b8d5ad9d974a44e7e2882f986467f4d3∗ [email protected]

Abstract for decisions by funders and recruitment com- mittees. The current system of scholarly communica- There are many problems with the current tion is based on tradition, and does not corre- state of affairs, including spond to the requirements of modern research. The dissemination of scientific results is cost Scientific output is to a large extent con- mostly done in the form of conventional arti- trolled by a profit-driven in- cles in scientific journals, and has not evolved dustry that provide little added value. In with research practice. fact, many scientific journals only consist In this paper, we propose a system of aca- of collections of papers produced by pub- demic endorsement based on blockchain tech- lic funds, refereed for free by the scientific nology that is decoupled from the publication community. The role of the journal is then process, which will allow expeditious appraisal simply to provide a publishing platform, of all kinds of scientific output in a transparent organize the refereeing process and make manner without relying on any central author- editorial decisions. In spite of this, scien- ity. tists are expected to relinquish copyright for their articles and pay the journal for publishing their work. In addition to this, Introduction many journals charge the scientific com- munity for accessing the articles, (unless Scientific output is traditionally disseminated exuberant fees are payed by in the form of articles in scientific journals. It the authors), thereby effectively prevent- is then given value by peers and funders based ing access from academics at universities on in which journal it is published, e.g. by without a journal subscription, and from the Journal , and by counting the public. Furthermore, these so-called the number of other works citing the paper. paywalls hinder meta-analytics, and drive In recent years, other metrics of scientific im- scientists to resort to services of question- pact have appeared, but they are rarely used able legality, such as SciHub[2]. ∗As our proposal aims to make parts of the scien- tific publishing industry obsolete, we have chosen to metrics In spite of common belief, the merit publish this proposal under a nom de plume in order to of an article is not automatically high be- minimize the risk of adverse effects (e.g. unfavourable cause it is published in a prestigious jour- editorial decisions, or stern letters to our employers[1]). The authors declare that they have no competing in- nal. The Journal Impact Factor (JIF) is terests. often used by funding agencies to rank ap-

1 plicants, in spite of the questionable value Proposal: an academic endorse- of this metric.[3, 4, 5, 6, 7]. ment system control Given the influence editorial decisions The purpose of evaluating research output is to can have on scientific careers, in particular guide scientists, funders, recruiters in making for young scientists, publishers and editors different kinds of decisions; what articles are are bestowed an undue power. The “cult worth spending time reading, how to fund the of the journal” is detrimental to scientific best research, which scientists are most likely progress[8]. In fact, the entire system of to produce important results. It should be pos- peer reviewing comes with a considerable sible to identify high-quality research and per- risk of preventing outstanding discoveries formant researchers without waiting for cita- to be published[9, 10]. tion data, as that is a slow process. In this communication, we propose an aca- delays Scientific projects often take years demic endorsement system (AES) with a form from start to publication. Consequently, of currency, academic endorsement points there is an inherent time lag in using cita- (AEP), which scientists can use to endorse re- tions to appraise scientific work. search output. Each scientist is then (possi- bly periodically) credited with AEP to reward publication bias There are very small incen- scientific work that he/she finds worthy of en- tives to publish negative results and stud- dorsement. The amount of AEP each scientist ies confirming previous results, as this is credited is based on the amount of AEP re- kind of work, while important for further- ceived for previous work. The total amount of ing science, is less likely to garner citations AEP given to a research object can then be in the future. used as a metric for its value. The amount of AEP given to a scientist can be used as a non-publications It is difficult to be re- measure of his/her impact. warded for non-traditional output, e.g. While impact and importance of scientific data sets and scientific software. work is not amenable to be quantified as simple numbers – metrics should always be analyzed While several of the deficiencies outlined in a context, the advantages for such a system above are being addressed in various projects are many-fold: and organizations, in particular new ways of scientific dissemination, a key feature that is • Any kind of scientific output could be en- missing in most of these initiatives is the pos- dorsed, as long as there is a persistent sibility to receive merit in other ways than identifier for the research object. This citations. Novel metrics for scientific impact would facilitate attributing value to non- are often based on journal publications[11], or traditional publications (e.g. arxive doc- more oriented towards providing reading sug- uments, open referee reports, blog posts, gestions[12, 13, 14]. data sets, software, etc.). The short-comings of current evaluation sys- tems are well-known in the scientific commu- • Provided there are clear links to the au- nity[3]; the San Francisco Declaration on Re- thors (e.g. as part of the PID metadata), search Assessment[15] has more than 104 sig- they would be credited AEP from endorse- natures at the time of writing. ments by their peers.

2 • With enough uptake, the value of new re- plored for e.g. governance systems and venture sults will become apparent much faster capital funds[20]. than with citation metrics. In the proposed AES, each scientist would then have an EOA, and every digital research • Scientists whose output has been endorsed object would have a smart contract attached to to a high degree will have more AEP to use it. An endorsement would be a transaction of for endorsements, and thus have a larger AEP from an EOA to a scientific object, where influence in the community. the amount of AEP transferred – limited by the endorsers current balance – and would reflect Transparency and decentralization by the importance the endorser is attributing the using blockchain technology work. EOA:s would thus only be able to trans- fer merit points to research objects (i.e. sci- The proposed system of endorsement would entific output, not scientists can be credited). need to be carefully designed in order to mini- Conversely, research objects (smart contracts) mize opportunities to game the system to gain would only transfer AEP to scientists. undue advantages. It would also be better The main advantages for using blockchains served by not having a single organization con- to track academic endorsements are that the trolling it. This is why we propose to use system can then be autonomous, and have blockchain technology as the underlying infras- complete transparency. The distributed ledger tructure for the AES. model will allow consumers of this informa- Blockchain technology is perhaps most well- tion to only take endorsements from a subset known for its use in crypto currencies, e.g. bit- of users (i.e. trusted scientists) into account, coin. However, in recent years, the technology if they so wish – thereby providing means to has evolved and is now being proposed for dif- circumvent attempts of gaming the AES. This ferent kind of point systems, tracking owner- filtering could be based on the amount of AEP ships, educational records[16], smart contracts, received, but also by identifying the persons decentralized name resolution[17] and even for attached to the accounts, i.e. allowing scien- generating persistent identifiers[18]. tists to follow what authorities in their field Briefly, blockchain technology allows for dis- endorse (and not only what they cite). In tributed verifications of transactions. Trans- addition, an endorsement system that is de- actions of tokens (e.g. crypto currency, votes, coupled from where scientific objects are pub- points) can take place between accounts. In lished/made available, will ameliorate the un- addition to accounts controlled by users of the sound dependency on the scientific publishing system (externally owned accounts, using the industry. Finally, the wealth of information terminology of the etherium platform[19]), ac- from such a system would allow for interesting counts can also be governed by predefined rules analytics on scientific progress. – so-called smart contracts. By using smart A prerequisite for the AES is to have unique contracts, it is thus possible to construct a sys- identifiers of research objects as well as of re- tem of transactions with its own set of rules searchers. In order to bootstrap the AES, the that, once constructed and deployed, can op- former could be built upon existing identifiers, erate autonomously, exempt from any form of e.g. the doi system[21], which is already used control by a central authority. These con- for articles, data sets and even software. The structs, known as distributed autonomous or- latter could be e.g. Open Researcher Id (OR- ganizations (DAOs), are currently being ex- CID)[22]. It would even be possible to boot-

3 strap the system by crediting initial users with a starting point for further discussions. There a pre-defined amount of AEP. The final re- are several fora where this discussion could quirement is an infrastructure where transac- take place; the OECD Directorate for Science, tions - based on the current rules - can be Technology and Innovation[25] have a strong recorded and verified. This could be con- record of furthering Open Science, as has the structed as a DAO on the ethereum blockchain European Union Directorate-General for Re- application platform[19, 23], but there are also search and Innovation[26]. On the grass-roots other alternatives (e.g. Openchain[24]). level, FORCE11 was formed with the aim “to The basic prerequisites would need to be bring about a change in modern scholarly com- complemented by an ecosystem of tools to fa- munications through the effective use of in- cilitate endorsements, viz. but- formation technology”[27], and the Research tons at article pages, and tools to analyze and Data Alliance[28] is a cross-disciplinary orga- visualize AEP transactions and account bal- nization which has backing from many fund- ances. ing organizations as well as technical exper- For a functional AES, there are many ad- tise among its members. Other organiza- ditional details to be discussed, e.g. how tions and projects include1 the W3C Research AEP is distributed when there are multiple au- Object for Scholarly Communication (ROSC) thors, how to handle retractions and scientific Community Group[29], OpenBlockchain[30], fraud, the possibilities of different endorsement CODATA[31], Pasteur4OA[32], researchob- flavours (e.g. novelty, quality), whether scien- ject.org[33], ePIC[34], FAIRDOM[35], the Dig- tists who contribute to the system by credit- ital Curation Centre[36] and the European ing scientific work should be rewarded in order Open Science Cloud[37]2 to incentivize its use, counteracting tit-for-tat Once these ideas have been scrutinized, dis- schemes and nepotism, the question of interest cussed and improved, we believe that the best rates on the AEP, etc. way forward is to convince funders about how an AES could improve research in general, and gain commitments from some large funding Discussion agencies to fund the development of a pilot sys- tem, and subsequently evaluate its usefulness In this paper we argue that a new way of giv- in ranking funding applications. ing merit to scientific results will accelerate A final word of caution: the scientific pub- scientific progress, and at the same time de- lishing industry will no doubt oppose any de- crease the scientific communitys dependence velopment that threaten their influence and on the publishing industry, which will free up profit margins3; their control of major commu- funds to research. The solution that we pro- nication channels combined with their lobby- pose – an academic endorsement system built ing experience will make this an uphill battle. on blockchain technology, could leverage ex- isting digital infrastructures, and would only 1Apologies if stakeholders are missing due to the au- need modest resources to realize. The major thors involuntary ignorance. challenges are achieving consensus for devising 2On a side note, the EOSC is proposing to award the AES, and ensuring uptake by the scientific cloud coins to be used by scientists when accessing IT community. resources – this concept would also be suited for crypto- currency technology. The purpose of this communication is to 3This is not out of malice, it is simply their respon- present the concept of a decentralized AES as sibility towards their shareholders.

4 Hopefully, science – with the help of academic [9] K. Siler, K. Lee, and L. Bero. “Measuring integrity and new technology – will prevail. the effectiveness of scientific gatekeep- ing”. In: Proc. Natl. Acad. Sci. U.S.A. References 112.2 (Jan. 2015), pp. 360–365. [10] Nope! 8 Rejected Papers That Won the [1] Letter from Association of Amer- Nobel Prize. url: : / / www . ican Publishers. url: https : / / authorea.com/users/8850/articles/ dl . dropboxusercontent . com / u / 117724/ (visited on 08/12/2016). 51751685/misc/ThomasAllen_Letter. [11] . “Eigenfactor: Measur- pdf (visited on 08/10/2016). ing the value and prestige of scholarly [2] Sci-Hub. url: http://sci-hub.cc/. journals”. In: College & Research Li- [3] Colin Steele, Linda Butler, and Danny braries 68.5 (2007), pp. 314–316. Kingsley. “The publishing imperative: eprint: http : / / crln . acrl . org / the pervasive influence of publication content / 68 / 5 / 314 . full . pdf + html. url metrics”. In: Learned Publishing 19.4 : http://crln.acrl.org/content/ (2006), pp. 277–290. issn: 1741-4857. 68/5/314.short. doi: 10 . 1087 / 095315106778690751. [12] Altmetric. url: https : / / www . url: http://dx.doi.org/10.1087/ altmetric.com/. 095315106778690751. [13] Mendeley. url: https : / / www . [4] P. O. Seglen. “Why the impact factor of mendeley.com/. journals should not be used for evaluat- [14] Authorea. url: https : / / www . ing research”. In: BMJ 314.7079 (Feb. authorea.com/. 1997), pp. 498–502. [15] San Francisco Declaration on Research [5] P. A. Lawrence. “The mismeasurement Assessment. url: http : / / www . ascb . of science”. In: Curr. Biol. 17.15 (Aug. org/dora/ (visited on 08/12/2016). 2007), R583–585. [16] Blockchains and the Web Position Pa- [6] P. A. Lawrence. “The Last 50 Years: per, A W3C Workshop on Distributed Mismeasurement and Mismanagement Ledgers on the Web. url: https : / / Are Impeding Scientific Research”. In: www . w3 . org / 2016 / 04 / blockchain - Curr. Top. Dev. Biol. 116 (2016), workshop/interest/third.html (vis- pp. 617–631. ited on 08/11/2016). [7] T. C. Ha, S. B. Tan, and K. C. Soo. “The [17] Namecoin. url: https : / / bit . journal impact factor: too much of an namecoin.org/. impact?” In: Ann. Acad. Med. Singap. 35.12 (Dec. 2006), pp. 911–916. [18] L. Bolikowski, A. Nowiski, and W. Syl- westrzak. “A System for Distributed [8] P. A. Lawrence. “The politics of publica- Minting and Management of Persistent tion”. In: Nature 422.6929 (Mar. 2003), Identifiers”. In: The International Jour- pp. 259–261. nal of Digital Curation 10.1 (2015), pp. 280–286. doi: 10 . 2218 / ijdc . v10i1.368.

5 [19] Ethereum. url: https : / / ethereum . [37] EOSC. url: http : / / ec . europa . eu / org/. research / openscience / index . cfm ? [20] The DAO. url: https://daohub.org/. pg=open-science-cloud. [21] DOI. url: http://www.doi.org/. [22] ORCID. url: https://orcid.org. [23] Ethereum . url: https:// . com / ethereum / wiki / wiki / White-Paper (visited on 08/11/2016). [24] Openchain. url: https : / / www . openchain.org/. [25] OECD Directorate for Science, Technol- ogy and Innovation. url: http://www. oecd.org/sti/. [26] EU Directorate-General for Research and Innovation. url: http : / / ec . europa.eu/research/index.cfm?pg= dg. [27] FORCE11. url: https://force11.org. [28] Research Data Alliance. url: https:// rd-alliance.org/. [29] W3C Research Object for Scholarly Communication Community Group. url: https://www.w3.org/community/ rosc/. [30] OpenBlockchain. url: http : / / blockchain.open.ac.uk/. [31] CODATA. url: http://www.codata. org/. [32] PASTEUR4OA. url: http : / / www . pasteur4oa.eu/. [33] researchobject.org. url: http : / / researchobject.org. [34] ePIC. url: http://www.epforum.eu/. [35] FAIR-dom. url: http : / / fair - dom . org. [36] The Digital Curation Centre. url: http: //www.dcc.ac.uk/.

6